Atom Paged Feeds Extension

[Robert Sanderson]

Dear All,

Foresite's primary scenario is creating resource maps for journals held by JSTOR (www.jstor.org), some of which span centuries of publication.  With all of the information we can include about an issue, a single resource map document that has every issue would be unusably large, so we need some sort of paging extension.

There's RFC 5005  (http://www.ietf.org/rfc/rfc5005.txt) which proposes the use of next/previous/first/last link relations in atom, and seems to be the best candidate.

There are two main issues with paging:

* We have a graph model underneath which is inherently unordered.  Paging through a graph makes no sense, so we need some way to know which page a resource is on.  Until such time as consensus is reached, I propose a simple orex:page predicate which takes a literal integer as object.
eg  my-photo-1 orex:page 1 ; my-photo-100 orex:page 6 ; ...

* We need a way to ask for a page from a resource map.  In python at least I think the easiest way to do this is add an optional 'page' argument to the serialisation methods.  -1 would be the default and mean all pages (as opposed to the last page, perhaps too unpythonic?)  Then the resulting ReMDoc would be just the page requested, rather than the entire thing.

Opinions?


Rob